Category Classification of Text Data with Machine Learning Technique for Visualizing Flow of Conversation in Counseling
Category Classification of Text Data with Machine Learning Technique for Visualizing Flow of Conversation in Counseling
复制标题
利用机器学习技术对文本数据进行类别分类，实现咨询中对话流程的可视化
DOI：
10.1109/nicoint.2017.35
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
Yuma Hayashida ; Tomoya Uetsuji ; Yasuo Ebara ; Koji Koyamada
中科院分区：
文献类型：
--
作者：
Yamamoto Yusaku;Oksa Gabriel;Vajtersic Marian;Yuma Hayashida ; Tomoya Uetsuji ; Yasuo Ebara ; Koji Koyamada
初级咨询师更倾向于为自己的利益继续咨询，他们更倾向于使用封闭式问题来与来访者确认解释。虽然专家咨询师正在指导咨询技巧初学者咨询师，我们认为客户对初学者咨询师的问题的反应是重要的，以适当的方法可视化。为了回应这一要求，我们开发了一个系统，用于可视化咨询中的会话流。然而，由于目前系统中会话数据的类别分类准确率很低，作为系统用户的专家咨询师需要手动修正初始分类结果，工作负担较大。为了改善这个问题，我们已经实现了与SVM（支持向量机）作为机器学习技术的文本数据的类别分类方法，以可视化的会话流在咨询。此外，我们还与现有系统的初始分类方法的结果进行了比较和评估。这些结果表明，使用SVM的分类方法的准确率比当前系统中的结果更高。
The beginner counselors have more likely to continue counseling in their own interest, they have a high tendency to make great use of the closed-ended question in order to confirm the interpretation with the client. While expert counselors are instructing the counseling skill to beginner counselors, we consider that the reaction of a client for a beginner counselor's question is important to visualize in an appropriate method. To respond the request, we have developed a system for visualizing the flow of conversation in counseling. However, the expert counselor as the system user requires to correct the initial classification result manually, and the work burden is large, because the accuracy of the category classification of conversation data is very low in the current system. To improve this problem, we have implemented on the category classification method of text data with SVM (Support Vector Machine) as machine learning technique to visualize the flow of conversation in counseling. In addition, we have compared and evaluated with results of the initial classification method of the current system. As these results, we have shown that the accuracy rate of the classification method with SVM become higher than the results in the current system.